VIP Call Girl Service Andheri West ⚡ 9920725232 What It Takes To Be The Best ...
Qbt npl recovery system prediction utool_en_2014
1. PREVISIONI DI INCASSO NEL SETTORE NPL
NPL Recovery system prediction
Made in Switzerland
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
2. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
The
idea
The economic crisis has lowered the percentage of recovery and has
made borrowers more consistent - in the sense that it has become more
difficult to distinguish between borrowers who pay for those who do not
pay again their loans already non performing.
Make predictions has thus become much more difficult and QBT, in 2012,
gained the conviction of the need to develop new algorithms and
methods of credit analysis.
Utool is a collaboration between QBT and the Dalle Molle Institute for
Artificial Intelligence Studies, with funding obtained from the Swiss
Confederation through the Commission for Technology and Innovation
KTI.
UTool is a software which allows, given a set of data on loans in litigation,
to determine an expected recovery at the portfolio level and the measure
of the probability of recovery for each title.
3. PREVISIONI DI INCASSO NEL SETTORE NPL
As the Confederation's innovation promotion agency, CTI
lends support to R&D projects, to entrepre-neurship as well
as to the development of start-up companies. CTI helps to
optimise knowledge and technology transfer through the use
of national thematic networks.
Support is generally available for R&D projects relating to scientific innovations in all disciplines. Project
proposals are submitted using the bottom-up principle and are mainly selected on the basis of their
innovativeness and market potential.
The Swiss AI Lab IDSIA (Istituto
Dalle Molle di Studi
sull'Intelligenza Artificiale) is a
non-profit oriented research
institute for artificial intelligence.
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Partners
of
the
ini/a/ve
Università della Svizzera italiana
Scuola universitaria professionale
della Svizzera italiana
IDSIA
Istituto Dalle Molle di studi
sull’intelligenza artificiale
The Swiss AI Lab IDSIA (Istituto Dalle Molle di Studi sull'Intelligenza Artificiale) is a non-profit oriented
research institute for artificial intelligence, affiliated with both the Faculty of Informatics of the Università della
Svizzera Italiana and the Department of Innovative Technologies of SUPSI, the University of Applied
Sciences of Southern Switzerland. We focus on machine learning (artificial neural networks, reinforcement
learning), optimal universal artificial intelligence and optimal rational agents, operations research, complexity
theory, and robotics. IDSIA is situated near Lugano, a lakeside city in the Italian-speaking canton of Ticino, a
region of Switzerland well known for its warm climate and outstanding scenery.
4. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Technology
To evaluate a portfolio of NPL means to quantify the prediction of the cash flow(forecast collection).
We can identify three main phases:
Acquisition: for estimating the cash flows from a portfolio of loans to determine the purchase price.
Management:
Prediction of the future collections of the entire portfolio and/or loan by loan.
Determination of the best action to recover the debt (phone collection, exactions, etc.): scenario analysis.
Assignment: selection of the borrowers particularly problematic in order to sell them to third parties.
In general it is not possible to precisely determine the value of a specific npl and therefore the portfolio, but
you can only give probabilistic assessments of trying to frame the particular debtor in a class of risk.
For this reason, in the three phases mentioned above we talk about estimation and prediction because of
the fact that information that are available and uncertainty, do not allow to exactly determine the value of a
loan or a borrower but only to give an expected value. Uncertainty that is normally present in the
enhancement of receivables in dispute was recently added to the uncertainty arising from economic crisis
that introduced further variability and therefore made unreliable the previous methods of credit evaluation.
5. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Technology
The new method uses the informations available on the nature of the loan and of the
borrower (typically collected during the due dilgience) to learn the behavior (model) of sundry
borrowers as well as statistical information, even dynamic (changing over time) gathered
during the activity of management of portfolios.
We can also add informations about the liquidity of the borrower (when available), the ability
of pressure on him (leverage, eg. guarantees), or information about external factors such as
macro-economic (GDP, unemployment rate, etc.), to improve the accuracy of the estimate /
forecast taking into account the general economy.
The algorithm that we have developed
allows for reliable predictions even on
recent portfolios (2010-2012), in which the
algorithms developed before the economic
crisis estimate too optimistic (i.e.
overestimate the percentage of
collections).
To implement these models we used
statistical techniques and algorithms of the
most advanced (which in particular allow to
make reliable predictions in conditions of
strong uncertainty http: //ipg.idsia.ch/)
developed by IDSIA over the past 20 years.
6. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Evalua/on
sampling
Consider the case of an actual portfolio of unsecured loans that has to be acquired by a
investor.
Suppose that investor has the historical collections of the portfolios currently under
management of which are known the following features:
Date of default
Date of beginning of the legal procedure, if any
Geographical attribute for the borrower (province, region, nation)
Private/Company
Availability of borrower
City class (size in inhabitants of the town)
Socio-economic classification by statistical research center (ISTAT)
Phase (still in progress, with enclosed court, judicial closed etc.).
Type of loan/borrower (e.g. classification management .: small - medium - big)
Age of the borrower
Ø GBV or gross book value
Ø Income
This information are the INPUT of "training" of our system, which is then applied to the NPL
portfolio that has to be valued.
Note: Each servicer or investor in NPL has these or similar information regarding its historical activities so
UTool is then automatically and implicitly customized for each client.
7. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Evalua/on
sampling
Suppose that we want to evaluate a portfolio that does not contain all but only some of this
information:
√ Date of default
Date of beginning of the legal procedure, if any
√ Geographical attribute for the borrower (province, region, nation)
√ Private/Company
Availability of borrower
√ City class (size in inhabitants of the town)
√ Socio-economic classification by statistical research center (ISTAT)
√ Phase (still in progress, with enclosed court, judicial closed etc.).
√ Type of loan/borrower (e.g. classification management .: small - medium - big)
Age of the borrower
à GBV or gross book value
à Income: this variable becomes the OUTPUT of our prediction on credit recovery
8. PREVISIONI DI INCASSO NEL SETTORE NPL
Evalua/on
sampling:
a
back
test
To confirm the validity of our approach we performed a series of
calculations based on data from the past to obtain testable predictions.
Below are the results for a portfolio of 5,000 loans with a series of
historical data for 20,000 borrowers.
GBV Pedicted income Pred. Inc Vs GBV Real income Real Inc. Vs GBV Delta %
312.538.905,06 28.287.401,20 9,05% 30.090.831,50 9,63% 0,58%
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
9. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
How
to
use
UTool
The Utool system can be proivded to our customer as a product or as a
service:
10. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Output
example
Here below you can find an example of the output of UTool in CSV file
format:
id Worst Average Best Rank Info Rank
1 2.464,37 7 .659,13 35.736,88 168 34
2 23,45 73,45 333,57 247 136
3 299,65 926,22 4.318,08 221 91
4 1,21 6,20 16,72 256 173
5 23,45 73,45 333,57 247 136
6 334,00 910,22 4.353,53 220 90
7 0,03 0,10 0,40 273 253
8 2.245,22 5 .416,18 33.913,13 168 34
9 0,02 0,06 0,38 273 253
16 1.137,27 5 .115,97 41.366,87 17 257
17 2.245,22 5 .416,18 33.913,13 168 34
18 260,38 736,64 3.667,77 221 91
Ø Where each line is a loan or a borrower:
Ø “ID” is the NDG of the loan/borrower
Ø “Worst” is the prudential value of the prediction of the income
Ø “Average” is the correct estimation …
Ø “Best” is the aggressive value …
Ø “Rank” : if higher it means that the borrower has a good attitude to re-pay its debt
Ø “Info Rank” : if higher, more information are needed to perform a better prediction à
optimal for sampling for due-diligence
11. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Group
Overview
QBT
Finance
Financial
Solu@on
QBT
NLP
Natural
Language
Processing
Page
11
QBT
Sagl
Founded
in
2008
in
Chiasso
QBT
SoHware
SoGware
Development
12. PREVISIONI DI INCASSO NEL SETTORE NPL
Ø QBT
stands
for
Quantum
Bit
Technology:
we
promote
the
idea
of
a
Company
opera@ng
in
the
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
field
of
applied
science.
Ø Established
in
year
2008,
QBT
is
ac@ve
both
in
research
and
commercial
projects
Ø The
mission
of
the
Company
is
to
develop
algorithms
and
related
soGware
/
hardware
on
behalf
of
our
Customers
or
on
our
own
for
the
projects
in
which
we
believe.
Ø QBT
offers
three
core
services
to
financial
and
non
financial
company:
•
Natural
Language
Processing
across
different
sectors
(finance,
law,
human
resources…):
ü Seman@c
Technology
(Machine
Learning)
ü Web
crawling
(we
customize
web
crawls
to
extract
data
from
websites)
ü Agent
based
models
to
reproduce
in
a
laboratory
a
simula@on
of
social
scenarios
•
Financial
solu@ons
(data
warehouse,
por]olio
analysis,
ra@ng,
due
diligence
systems)
especially
for
the
NPL
and
Real
Estate
sectors
•
SoGware
development
from
analysis
to
code-‐wri@ng
Ø More
then
20%
of
profit
invested
in
R&D
Pa
ge
12
Philosophy
13. PREVISIONI DI INCASSO NEL SETTORE NPL
Newsmark
et
2012 2013
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Page
13
NRGLEX
Chiasso
–
www.qbt.ch
Services
and
products
Time
NPL
Logistics
Real Estate
2008 2009
Public
Institution
2010
Software development
per sector / product
R&D activities
NPL -
IDSIA ABM - CNR NLP - CNR
Sentiment
Balanscehe
et Rating
Ontonix
RAS
Complexity
UTOOL
Oracolo
Urbano
HR
Sentiment
TODY
2014
NLP NLP - USI
14. PREVISIONI DI INCASSO NEL SETTORE NPL
Partners
&
Customers
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Page
14
15. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT Sagl informs the addressees of this document that all the information
contained therein are confidential and can not be (i) copied, (ii) disclosed to third
parties and (iii) disclosed in any form, without the prior permission of QBT Sagl .
This information can not have any contractual force because they are supplied
only for commercial and promotional purposes relating shares of the fund to
which they relate and can not, therefore, be used for other purposes.
Any violation set forth above will be prosecuted according to law.
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Disclaimer
Page
15
16. PREVISIONI DI INCASSO NEL SETTORE NPL
QBT
Sagl
–
Via
E.
Bossi
4,
6830
Chiasso
–
www.qbt.ch
Thank
You
QBT
Sagl
Via
E.Bossi,
4
CH-‐6830
Chiasso
–
Svizzera
web:
www.qbt.ch
–
email:
info@qbt.ch
Tel.:
0041
(0)
91.682.24.28
Fax:
0041
(0)
91.682.11.21
Contact
Informa/on